| Classification of Eye States Using EEG Data: A Comparative Analysis of Machine Learning Algorithms |
| Paper ID : 1009-ICEEM2025 (R2) |
| Authors |
|
Marwa M. Eid1, Amal H. Alharbi2, Amel Ali Alhussan2, S.K. Towfek *3, Farah Adlan4, Doaa Sami Khafaga2 1Faculty of Artificial Intelligence Delta University for Science and Technology Mansoura 35111, Egypt 2Department of Computer Sciences College of Computer and Information Sciences Princess Nourah bint Abdulrahman University Riyadh, Saudi Arabia 3Computer Science and Intelligent Systems Research Center, Blacksburg 24060, Virginia, USA 4Department of Communications and Electronics Delta Higher Institute of Engineering and Technology Mansoura, Egypt |
| Abstract |
| Classification of close or open-eye states using EEG data is slightly problematic since the EEG signals are complex. Well-developed classification techniques can enhance the functions of disease diagnosis, man-machine interfaces, and aids for people with disabilities. For this purpose, we utilized an annotated EEG dataset, labeled two states, eye-open (state 0 ) and eye-closed (state 1 ). For analysis, we used KNeighborsClassifier, RandomForestClassifier, DecisionTreeClassifier, LogisticRegres-sion, and GaussianNB classifiers for their ability to determine and predict the given eye states. The accuracy, sensitivity, specificity, and other usable parameters that defined each model’s efficacy were used to gauge its effectiveness. Therefore, it shows a good classification ability with the KNeighborsClassifier model, which has 95.86% accuracy among all the tested models. This was followed by the Random module’s classifier with a 91.9% accuracy, demonstrating another candidacy. Yet, more basic algorithms like LogisticRegression and GaussianNB indicated relatively worse performance and thus lower ability to accommodate specific features of EEG signals. Our findings emphasize the importance of a model selection strategy in achieving high accuracy for eye-state classification. The analysis can filter machine learning models appropriate for EEG where accurate state detection is critical. |
| Keywords |
| Eye state classification, EEG, machine learning, Random Forest, Classifier, accuracy |
| Status: Accepted |